Fuel 90 (2011) 3461–3467

Contents lists available at ScienceDirect

Fuel

journal homepage: www.elsevier .com/locate / fuel

Towards continuous sustainable processes for enzymatic synthesis of biodieselin hydrophobic ionic liquids/supercritical carbon dioxide biphasic systems

Pedro Lozano a,⇑, Juana M. Bernal a,1, Michel Vaultier b,2

a Departamento de Bioquímica y Biología Molecular B e Inmunología, Facultad de Química, Universidad de Murcia, P.O. Box 4021, E-30100 Murcia, Spainb Laboratoire de Chimie et Photonique Moléculaires, UMR CNRS 6510, Université de Rennes-1, Campus de Beaulieu, Av. Général Leclerc, F-35042 Rennes, France

a r t i c l e i n f o a b s t r a c t

Article history:Received 11 May 2011Received in revised form 6 June 2011Accepted 8 June 2011Available online 30 June 2011

Keywords:BiodieselBiocatalysisSupercriticalGreen processContinuous reactor

0016-2361/$ - see front matter � 2011 Elsevier Ltd. Adoi:10.1016/j.fuel.2011.06.008

⇑ Corresponding author. Tel.: +34 868 88 73 92; faxE-mail address: plozanor@um.es (P. Lozano).

1 Tel.: +34 868 88 73 92; fax: +34 868 88 41 48.2 Tel.: +33 2 99 28 62 74; fax: +33 2 99 28 69 55.

The excellent suitability of immobilized Candida antarctica lipase B (Novozym 435) catalyst to carry outthe synthesis of methyl oleate (biodiesel) by methanolysis of triolein in ILs based on imidazolium cationswith large alkyl side chain (from C12 to C18) has been demonstrated at 60 and 85 �C. The phase behaviourof IL/triolein/methanol and IL/methyl oleate mixtures were studied at different concentrations and tem-peratures, the best results (up to 98.6% biodiesel yield after 6 h) being obtained for ILs able to providemonophasic reaction systems, i.e. 1-methyl-3-octadecylimidazolium bis(trifluoromethylsulfonyl)imide).A continuous enzymatic reactor, based on biocatalysts particles coated with hydrophobic ILs, for biodie-sel synthesis in supercritical carbon dioxide was studied at 60 �C and 180 bar. The operational stability ofthe immobilized lipase was improved by its coating with ILs, i.e. 1-methyl-3-octadecylimidazolium hexa-fluorophosphate, leading to a two-phase systems with respect to the biodiesel product, which showed anexcellent catalytic behaviour in continuous operation under supercritical conditions (up to 82% biodieselyield after 12 cycles of 4 h).

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Increased demand for energy, global warming due to emissionof green house gases, environmental pollution, and fast dimin-ishing supply of fossil fuels are the major key factors leadingto search for alternative sources of energy [1]. Biodiesel is a die-sel substitute fuel, composed of fatty acid methyl esters (i.e.methyl oleate) and obtained from renewable sources (i.e. vegeta-ble oils, fats, etc) by catalytic transesterification with primaryaliphatic alcohols (e.g. methanol). The interest to produce biodie-sel by clean and sustainable approaches including biocatalysis isdoubtless, because of the improved quality of the final product,as well as the reduction in wastewater production during sepa-ration and purification steps [2]. At the opposite of chemical cat-alysts, biocatalysts (i.e. lipases) allow the synthesis of specificalkyl esters, the easy recovery of glycerol and the transesterifica-tion of fat substrates with high free fatty acid content [3].However, one of the common drawbacks with the use ofenzyme-based processes is the high cost of the enzyme, makingit necessary to develop reusable biocatalyst derivatives with highoperational stability. In this context, the immiscibility of triacyl-glycerides (e.g. triolein) and alcohols (e.g. methanol) clearly

ll rights reserved.

: +34 868 88 41 48.

impairs their exploitation at industrial scale [4,5]. These two-phase reaction systems exhibit a decrease in the catalytic effi-ciency of the enzyme because of its fast deactivation resultingfrom its direct contact with methanol, as well as, the low effi-ciency of substrates transport towards the enzyme microenvi-ronment. All these features result in a decrease of both theturnover frequency and the number of recycling operations ofthe biocatalysts [6].

Methanol stepwise addition, acyl acceptor alterations and/orsolvent engineering are the applied approaches to solve all theseconstrains. The stepwise addition of methanol [7,8], or the additionof silica particles containing adsorbed methanol [9], permit up to94% biodiesel yield in a 24 h period, and the enzyme reuse for sixcycles as well. When methanol is replaced by a different acylacceptor, like methyl acetate, the lipase inactivation by methanolinteraction is reduced, resulting in up to 92% biodiesel yield in24 h for a 12:1 methyl acetate:oil molar ratio [10]. Medium engi-neering approaches to improve methanol solubility into the reac-tion medium provided good results. Thus, some organic solvents(e.g. tert-butanol) improve methanol solubility into the vegetableoil thus avoiding the direct interaction between the enzyme andpure methanol. This results in the improvement of the biocatalyticefficiency (up to 97% biodiesel yield after 24 h at 50 �C) [11,12].Supercritical carbon dioxide (scCO2) could be a good choice as agreen solvent for enzymatic synthesis of biodiesel, but moderateyields (40–90%) and low operational stability of the enzyme havebeen reported [13–15].

OOC

OOC

OOC

OH

OOC

OOC

OH

OOC

OH

COO-CH3

OH

OH

OH

scCO2

Ionic Liquid

1

2Novozym COO-CH3

COO-CH3

B

n = 1, [C mim] n = 6, [C mim] n = 7, [C mim]

n = 5, [C mim] n = 8, [C mim]

14 4

16

12 18

n = 3, [C mim]8

NCH3(

n

[ ]BF4[ ]PF6

F C3 SO

O

N SO

OCF3

NTf2

A

) N

Fig. 1. A. Scheme of immobilized lipase-catalyzed synthesis of biodiesel (2, methyloleate) by methanolysis of triolein (1) in IL/scCO2 biphasic systems. B. Cations andanions of the assayed ILs.

3462 P. Lozano et al. / Fuel 90 (2011) 3461–3467

The biocatalytic synthesis of biodiesel in ILs has also been re-ported [16–19]. However, when the assayed ILs were based onshort-chain 1,3-dialkylimidazolium cations (e.g. [Bmim][PF6],[Bmim][NTf2]), as they were not able to dissolve triacylglycerides,the resulting two-phase reaction media only provide moderatebiocatalytic efficiency (up to 90% yield for 24 h at 60 �C) [17]. Theenzymatic synthesis of biodiesel performed using [Bmim][BF4]and [Bmim][PF6] gave similar conversions to that obtained in[Bmim][NTf2]. However, the hydrophobic [NTf2] anion seems tobe the best choice, since [BF4] is hydrophilic thus rending more dif-ficult the glycerol separation, while [PF6] may hydrolyze and gen-erate HF [18]. However, the better suitability of [Bmim][BF4] forlipase-catalyzed epoxidation of methyl oleate with H2O2 was re-cently reported with respect to [Bmim][PF6] and [Bmim][NTf2][20]. In the same way, water-miscible ILs based on hydrophilic an-ions (e.g. methylsulfate) and quaternary ammonium cations withlarge alkyl chains containing ether functional groups were shownto be excellent reaction media for lipase-catalyzed glycerolysis oftriglycerides [21]. In fact the rule for enzyme activity in ILs is notknown at the moment and seems to be ‘‘there is no rule’’, since per-formance in a particular IL appears to vary significantly from en-zyme to enzyme and reaction to reaction [22].

Recently its has been reported how hydrophobic ILs based oncations with large alkyl side-chain (e.g. 1-methyl-3-octadecylimi-dazolium bis(trifluoromethylsulfonyl)imide, [C18mim][NTf2]) wereable to dissolve both triolein and methanol, providing one-phasereaction media that showed an excellent suitability for the biocata-lytic synthesis of biodiesel, i.e. up to 96% yield in 6 h at 60 �C. Theenzyme/IL system was reused for 7-times without activity loss[23–25].

The most interesting feature for biocatalytic processes in ILs isthe possibility to design two-phase reaction systems that easily per-mit product recovery [26,27]. In this context, biocatalysis in IL/scCO2 biphasic systems has emerged as an exceptionally interestingapproach for designing continuous clean processes that directlyprovide pure products, because the advantages of scCO2 for masstransport are complemented by the high catalytic efficiency of en-zymes in ILs [28]. This paper shows for the first time the biocatalyticsynthesis of biodiesel in IL/scCO2 biphasic systems (see Fig. 1). Thus,the phase behaviour of IL/triolein/methanol and IL/methyl oleatemixtures, as well as its influence for lipase-catalyzed biodiesel syn-thesis in ILs and IL/scCO2 reaction media has been studied. Follow-ing the statement like-dissolves-like, twelve different ILs based onimidazolium cations with a large alkyl side chain (from C12 to C18)and [BF4], [PF6] or [NTf2] anions were assayed. The obtained resultsclearly demonstrate the suitability of IL/scCO2 biphasic systems forthe continuous enzymatic synthesis of biodiesel as a sustainableprocess.

2. Experimental

2.1. Materials

Immobilized Candida antarctica lipase B (Novozym 435�, EC3.1.1.3) was a gift from Novozymes S.A. (Spain). Triolein (65% pur-ity), solvents and other chemicals were purchased from Sigma–Al-drich–Fluka (Madrid, Spain). The ILs (99% purity), 1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([C4mim][NTf2]), 1-octyl-3-methylimidazolium bis(trifluoromethylsulfo-nyl)imide ([C8mim][NTf2]), 1-dodecyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([C12mim][NTf2]), 1-tetradecyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([C14mim][NTf2], 1-hexadecyl-3-methylimidazolium bis(trifluoromethylsul-fonyl)imide ([C16mim][NTf2], 1-octadecyl-3-methylimidazoliumbis(trifluoromethylsulfonyl)imide ([C18mim][NTf2]), 1-dodecyl-3-methyl-imidazolium hexafluorophosphate ([C12mim][PF6]), 1-tet-

radecyl-3-methylimidazolium hexafluorophosphate ([C14mim][PF6], 1-hexadecyl-3-methylimidazolium hexafluorophosphate([C16mim][PF6], 1-octadecyl-3-methylimidazolium hexafluoro-phosphate ([C18mim][PF6]), 1-dodecyl-3-methylimidazolium tetra-fluoroborate ([C12mim] [BF4]), 1-tetradecyl-3-methylimidazoliumtetrafluoroborate ([C14mim][BF4], 1-hexadecyl-3-methylimidazo-lium tetrafluoroborate ([C16mim][BF4], and 1-octadecyl-3-methyl-imidazolium tetrafluoroborate ([C18mim][BF4]), were obtainedfrom IoLiTec GmbH (Germany). Melting points of ILs were deter-mined by using a Reichert Thermovar melting point apparatusequipped with a microscope, and were uncorrected.

2.2. Studies on phase behaviour of IL/substrates and IL/biodieselmixtures

Different amounts (0.09, 0.18, 0.29, 0.44, 0.60 and 0.67 g, respec-tively) of IL ([C12mim][NTf2], [C14mim][NTf2], [C16mim][NTf2] [C18mim][NTf2], [C12mim][PF6], [C14mim] [PF6], [C16mim][PF6], [C18mim][PF6], [C12mim][BF4], [C14mim][BF4], [C16mim][BF4]or [C18mim][BF4]) were added into six different screw-capped vialswith teflon-lined septa (1.5-mL total capacity). Then, variableamounts of methanol (0.16, 0.14, 0.12, 0.09, 0.06, and 0.04 g) wererespectively dissolved into the IL. Finally, the correspondingamounts of triolein were added to each tube in order to obtainthe following IL/methanol/triolein ternary mixtures (w/w/w): a,9.1/16.2/74.7, b, 18.5/14.6/66.9; c, 31.0/12.3/56.7; d, 47.6/9.4/43.0; e, 63.8/6.5/29.7; f, 73.7/4.2/22.1 (see Fig. 3 and Electronic Sup-plementary Information, ESI). In all cases, the triolein: methanol ra-tio was 1:6 (mol:mol). The resulting mixtures were incubated withshaking for 1 h at 60 �C or 85 �C, as a function of the type of anion.For the case of methyl oleate (biodiesel), the following IL/methyloleate binary mixtures (w/w) were assayed (1 g total amount): a,10/90; b, 20/80; c, 33/67; d, 50/50; e, 66/34 and f, 75/25 (See ESI).

2.3. Lipase-catalyzed biodiesel synthesis in ILs media

For each IL, triolein (0.36, 0.23 or 0.11 mmol) was added intothree different screw-capped vials (0.9-mL total capacity),

Filter

OVEN

Temperature

PressureMeOH

Triolein

Novozymcoated with IL

Liqu

id C

O T

ank

2 SyringePump

HeatedRestrictor

BIODIESEL

HPLC Pumps

Fig. 2. Experimental set-up of the continuous packed bed reactor containingNovozym 435 coated with IL for biodiesel synthesis by methanolysis of triolein.

Fig. 3. Phase behaviour of [1-alkyl-3-methylimidazolium][NTf2]/triolein/methanolmixtures at 60 �C. The following ratios were assayed: a, 9.1/16.2/74.7, b, 18.5/14.6/66.9; c, 31.0/12.3/56.7; d, 47.6/9.4/43.0; e, 63.8/6.5/29.7; f, 73.7/4.2/22.1 (w/w/w),respectively. For all cases, the [triolein]:[methanol] ratio was 1:6 (mol:mol). Thefollowing ILs are shown: A, [C8mim][NTf2]; B, [C12mim][NTf2]; C, [C14mim][NTf2]; D,[C16mim][NTf2]; E, [C18mim][NTf2].

P. Lozano et al. / Fuel 90 (2011) 3461–3467 3463

containing 77, 224 or 348 mg of IL, and methanol (2.18, 1.37 or0.69 mmol), respectively. The resulting final mixtures gave thefollowing IL/triolein/methanol ratios (w/w/w): 16.4/68.7/14.9;47.7/43.0/9.3; 73.9/21.4/4.7, respectively. For each case, the mix-ture was incubated in a thermoshaker TS-100 (Baeco, Germany)at the selected temperature and at 1000 rpm for 30 min. Thereaction was started by adding 10% (w/w) Novozym 435 with re-spect to the triolein amount, and the reaction system was incu-bated at 60 or 85 �C for 8 h. Then, a 20 lL aliquot was taken andadded to 480 lL dodecane/isopropanol (95:5, v/v) mixture, andthe resulting biphasic mixture was strongly shaken for 3 minto extract biodiesel. The resulting mixture was centrifuged at15,000 rpm for 10 min. Finally, 350 lL of dodecane/isopropanolextracts were added to 150 lL of 100 mM ethyl decanoate and100 mM tributyrin (internal standards) solution in dodecane/iso-propanol (95:5, v/v), and the final solution was analyzed by CG.

2.4. Lipase-catalyzed biodiesel synthesis in IL/scCO2 systems

Novozym 435 (1 g) was coated with 0.5 g of IL ([C14mim][NTf2],[C14mim][PF6], [C18mim][NTf2], [C18mim] [PF6] or [C18mim][BF4].respectively) as described previously in detail [29]. Novozym-ILparticles were placed in the cartridge of an ISCO 220SX (TeledyneIsco, Inc, Lincoln, NE, USA) high pressure extraction apparatus. Thisextractor is equipped with a syringe pump (ISCO 100DX, 100 mLoverall volume), and devices for pressure, temperature and flowrate control. The ISCO system was started by the continuous pump-ing of scCO2 at 180 bar and 60 �C, bubbling continuously CO2

through a calibrated heated restrictor (1 mL/min, 80 �C). Biodieselsynthesis was carried out for 4 h cycles by continuously pumpingtriolein (5.1 or 10.2 lmol/min mass flow rate) and methanol(244 lmol/min mass-flow rate) into the scCO2 inlet flow by usingtwo HPLC pumps (LC-10AT, Shimadzu Europe, Germany) (seeFig. 2). Substrates were transported by the scCO2 flow throughthe catalytic cartridge for biotransformation, and then productswere recovered by depressurizing through the calibrated heatedrestrictor for 30 min steps in a controlled amount of hexane/iso-propanol (5:4, v/v) placed on an ice-bath. Samples were analyzedby GC. In all cases, substrate and product mass-balances from theoutlet were consistent with the substrate mass-flow inlet.

2.5. GC analysis

GC analysis was performed with a Shimadzu GC-2010 (ShimadzuEurope, Germany) equipped with an FID detector. Samples wereanalyzed on a TRB-5HT capillary column (10 m � 0.32 mm �

0.1 lm, Supelco), using both ethyl decanoate and tributyrin as inter-nal standards, under the following conditions: carrier gas (He) at28.6 kPa (40 mL/min total flow); temperature programme 100 �C,10 �C/min, 200 �C, 15 �C/min, 370 �C, variable split ratio, (80:1 as10:1); detector, 220 �C. See additional information for peak reten-tion times and chromatograms.

3. Results and discussion

3.1. Phase behaviour of triolein/methanol/methyl oleate/hydrophobicIL systems

The suitability of twelve different ILs based on imidazolium cat-ions with one large alkyl side chain (from C12 to C18) to dissolveboth triolein and methanol substrates for biodiesel synthesis hasbeen studied. For each IL, the phase behaviour of six different IL/tri-olein/methanol ternary mixtures was studied at 60 or 85 �C,because of the solid nature of most assayed ILs at room tempera-ture (see Fig. 3 and ESI). In all cases the triolein/methanol ratiowas 1/6 (mol/mol), a proportion often considered as optimum forlipase-catalyzed biodiesel synthesis [16–18]. As it can be seen,

3464 P. Lozano et al. / Fuel 90 (2011) 3461–3467

the ability of these ILs to solubilise the triolein–methanol mixturewas clearly dependent on both the nature of ions and the IL con-tent. Thus, ILs based on the [NTf2] anion showed the best abilityto dissolve triolein, being improved by increasing both the lengthof the alkyl side chain of the cation and the IL concentration. Forexample, both [C16mim][NTf2] and [C18mim][NTf2] provided mon-ophasic systems for assayed concentration higher than 40% (w/w)(cases D and E, tubes d–f), while for the case of [C14mim][NTf2], thefull solubilisation of triolein only occurred when the IL concentra-tion was higher than 70% (w/w) (case C, tube f). By using ILs havingan alkyl chain in the cation shorter than C14, the full solubilisationof triolein was never observed. Similarly, the solubilisation of trio-lein/methanol mixtures in ILs based on [PF6] or [BF4] anion did notprovide one-phase systems in any of the assayed conditions, inde-pendently of the length of the alkyl side chain of cations, or even ifthe temperature was increased up to 85 �C, a temperature higherthan any melting point of the assayed ILs (see Table 1 and ESI).At this point, it is quite surprising how for [C16mim][PF6],[C18mim][PF6], [C14mim][BF4] [C16mim][BF4] and [C18mim][BF4]cases, the assayed mixtures with high IL content did not show afully liquid system at 85 �C, because of a remaining solid fractionof IL (see ESI). In fact, for the [C18mim][PF6] case, it was necessaryto heat up to 90 �C to reach a fully liquid media, while for all[C18mim][BF4] cases, the IL was maintained as solid even at

Table 1Phase behaviour of different IL/triolein or IL/methyl oleate mixtures as a function of the nNovozym 435 in each reaction medium after 8 h at 85 �C.

Ionic liquid Melting point (�C) IL (% w/w)

16.4[C12mim][NTf2] 17 47.7

73.9

16.4[C14mim][NTf2] 33 47.7

73.9

16.4[C16mim][NTf2] 46 47.7

73.9

16.4[C18mim][NTf2] 53 47.7

73.916.4

[C12mim][PF6] 58 47.773.916.4

[C14mim][PF6] 67 47.773.916.4

[C16mim][PF6] 74 47.773.916.4

[C18mim][PF6] 82 47.773.916.4

[C12mim][BF4] 30 47.773.916.4

[C14mim][BF4] 36 47.773.916.4

[C16mim][BF4] 49 47.773.916.4

[C18mim][BF4] 60 47.773.9

[C12mim]: 1-dodecyl-3-methylimidazolium; [C14mim]: 1-tetradecyl-3-methylimidazoliuylimidazolium; [PF6]: hexafluorophosphate; [BF4], tetrafluoroborate; [NTf2]: bis((trifluorL: monophasic liquid medium; L–L: liquid–liquid biphasic medium; S–L: solid–liquid bi

a IL/S phase: phases for IL/triolein/methanol mixtures.b IL/P phase. phases for IL/methyl oleate mixtures.

100 �C (results not included). For these resulting solid–liquid–li-quid systems at 85 �C, the bottom phase was essentially formedby solid IL and IL–methanol solution, while the upper phase mainlycontained triolein.

A similar study of miscibility between these hydrophobic ILsand both glycerol and methyl oleate was carried out. The glycerolby-product was always non-miscible with all the assayed hydro-phobic ILs, probably because of its high polar and hydrophilic char-acters. The phase behaviour of IL/methyl oleate binary systems wasalso studied under the same conditions as for the IL/triolein/meth-anol mixtures described in Fig. 3 (see ESI). As can be seen, the ILsbased on [NTf2] anion (from [C8mim] to [C18mim]) were fully mis-cible with methyl oleate under all the assayed conditions, provid-ing monophasic homogeneous systems at 60 �C. In the same way,most ILs based on [PF6] were also able to dissolve methyl oleate,but increasing temperature up to 85 �C, because of the high melt-ing point of these ILs. In this case, it is necessary to point out that[C18mim][PF6] was able to fully dissolve methyl oleate for all as-sayed mixtures at 85 �C, even though some solid IL was still pres-ent at this temperature. However, for ILs based on [BF4] anion, amonophasic system was only obtained for the most hydrophobiccation ([C18mim]) at low IL content. Furthermore, the increase inthe [C18mim][BF4] content prevents reaching the full liquid statein spite of the assayed temperature (85 �C), which was higher than

ature and concentration the IL, as well as the biodiesel yield produced by 10% (w/w)

IL/S phasesa IL/P phasesb Biodiesel yield (%)

L–L L 5L–L L 42L–L L 75

L–L L 6L–L L 42L L 72

L–L L 6L L 42L L 77

L L 7L L 43L L 79L–L L 6L–L L 32L–L L 41L–L L 8L–L L 33L–L L 36L–L L 9S–L–L L 36S–L–L S–L n.d.L–L S–L 9S–L–L S–L 37S–L–L S–L n.d.L–L L–L 6L–L L–L 6L–L L–L 26L–L L–L 7L–L L–L 12S–L–L S–L 18L–L L–L 4S–L–L L–L 5S–L–L S–L n.d.L–L L 5S–L–L L 5S–L–L S–L n.d.

m; [C16mim]: 1-hexadecyl-3-methylimidazolium; [C18mim]: 1-octadecyl-3-meth-omethyl)sulfonyl)imide.phasic medium; S–L–L: solid–liquid–liquid triphasic medium. n.d.: not detected.

P. Lozano et al. / Fuel 90 (2011) 3461–3467 3465

the melting point of this IL (60 �C). Similarly to the case of[C18mim][PF6], it seems that a fraction of the melted [C18mim][BF4]dissolves in the methyl oleate, while the remaining IL fraction staidin solid state. These results clearly show how both the triolein andthe methanol substrates for biodiesel synthesis can be homoge-neously distributed as a monophasic system, or heterogeneously,as a liquid–liquid or solid–liquid–liquid, multiphase systems, as afunction of both the nature of anion and the IL concentration.The resulting phase behaviour in substrates/IL and products/ILmixtures should clearly be involved in the performance of the bio-catalytic synthesis of biodiesel. In this context, as triolein is anhydrophobic compound which contains three C18 alkyl chains,these results suggest that the ability of any IL to dissolve trioleinmay be related to the increase in hydrophobicity of anion([NTf2] > [PF6] > [BF4]), as well as by the increase in length of the al-kyl chain in the cation, according to the statement like-dissolves-like. Thus, the suitability for lipase-catalyzed biodiesel synthesisin these systems should be assayed [23–25].

3.2. Lipase-catalyzed biodiesel synthesis in hydrophobic ILs

Fig. 4A depicts time courses of Novozym 435-catalyzed biodieselsynthesis in four different [1-alkyl-3-methylimidazolium][NTf2] ILsat 47.7% (w/w) IL concentration and 60 �C. As can be seen, the bio-diesel yield continuously increased with reaction time for all the as-sayed ILs, which pointed out its suitability as reaction media for theenzyme catalysis. However, the catalytic efficiency of the reactionsystem proportionally increased with the length of the alkyl chainof the cation (i.e. [C18mim] > [C12min] > [C8mim] > [C4mim], thebest results (up to 95% biodiesel yield in 8 h) being obtained forthe [C18mim][NTf2] case. These results could clearly be related with

Reaction time (h)0 2 4 6 8

Bio

dies

el Y

ield

(%)

0

20

40

60

80

100

Bio

dies

el Y

ield

(%)

0

20

40

60

80

100

[C 4mim

]

[C 8mim

]

[C 12mim

]

[C 14mim

]

[C 16mim

]

[C 18mim

]

A

B

Fig. 4. A. Time courses for the methanolysis of triolein catalyzed by Novozym 435in (N) [C18mim][NTf2], (�) [C12mim][NTf2], (.) [C8mim][NTf2], (j) [C4mim][NTf2]and (d) free solvent reaction media at 60 �C. B. Effect of the nature andconcentration of [1-alkyl-3-methylimidazolium][NTf2] ILs on the biodiesel yieldsynthesized by Novozym 435 at 60 �C.The following IL concentration were used:16.6 (white), 46.7 (dashed) and 73.7 w/w (black), respectively.

the miscibility of both triolein and methanol substrates with the as-sayed ILs (see Fig. 3). Thus, reaction mixtures shifted from a fullyclear monophasic medium (for the [C18mim][NTf2] case), towardsdifferent two phase reaction systems, where the size of trioleinlayer increased with decreasing the length of the alkyl chain of cat-ion. Furthermore, as the overall IL concentration has been shown askey parameter to achieve the full solubilisation of the reaction mix-ture (see Fig. 3), its influence on the catalytic efficiency of Novozym435 for biodiesel synthesis was also studied. Fig. 4B shows theresulting biodiesel yield after 8 h of reaction for six different [1-al-kyl-3-methylimidazolium][NTf2] ILs, as a function of the length ofthe alkyl chain of the cation, and for three different IL concentra-tions, i.e. 16.4, 46.7 and 73.7% w/w, respectively. The increase inthe IL concentration resulted in the enhancement in biodiesel yield,which improved by increasing the length of the alkyl side chain ofcation. As both [C4mim][NTf2] and [C8mim][NTf2] ILs were unableto dissolve triolein, the enzyme showed moderate activity for bio-diesel synthesis (up to 44% yield in 8 h) in all the resulting biphasicsystems. However, the enzyme was able to reach full triolein con-version in biodiesel at different [C14mim][NTf2], [C16mim][NTf2]and [C18mim][NTf2] contents. The higher IL content should clearlybe related with the better organisation of these nano-structuredreaction media, which allowed hydrophobic molecules (i.e. triolein)to reside in less polar regions, and polar species (i.e. methanol) toundergo faster diffusion in the more polar regions [30]. Thus, it alsowas reported how this extremely ordered supramolecular structureof ILs in liquid phase might also be able to act as a ‘‘mould’’, stabi-lizing the active 3-D structure of the enzyme in these non-aqueousnano-environments [28,31]. The decrease in the IL content couldalso be related with an enhancement of the direct interaction be-tween denaturing methanol molecules with the enzyme. In thesame context, the decrease in the hydrophobicity of the IL might in-volve a loss in the molecular organisation of these reaction mediacontaining hydrophobic substrates with large alkyl chains (e.g. tri-olein), that results in decreased mass-transport efficiency towardsthe enzyme microenvironment and so, in the catalytic efficiencyfor biodiesel synthesis.

The influence of the nature of anions (i.e. [NTf2], [PF6] and [BF4])of these hydrophobic ILs on the ability of the enzyme to catalyzebiodiesel synthesis was also studied. To compare results, the tem-perature was increased up to 85 �C to improve the melting/solu-bilisation of most of the assayed ILs, thus testing the biocatalyticreaction in liquid media (see ESI). Table 1 shows the phase behav-iour of the resulting IL/triolein/methanol (IL/S phase) and IL/methyloleate (IL/P phase) mixtures at three different IL content (i.e. 16.4,47.7 and 73.9 w/w, respectively), as well as the biodiesel yield pro-duced by the enzyme for each reaction system. As can be seen, bio-diesel yields in ILs based on [PF6] and [BF4] were clearly lower thanthose obtained for [NTf2] ILs, thus demonstrating the key role ofthe hydrophobicity of the anion ([NTf2] > [PF6] > [BF4] of these ILsto carry out this enzymatic reaction in liquid media. These resultsagain could be related with the phase behaviour of these reactionsystems, where heterogeneous multiphasic systems were shownin most cases. Thus, while the enzyme showed a low level activityin ILs based on [PF6] (<40% biodiesel yield in 8 h), it was practicallyinactive for ILs based on [BF4], including the highest IL concentra-tions. Once again, ILs based on the most hydrophobic anion (i.e.[NTf2]) provided the best results for the enzymatic synthesis of bio-diesel by increasing either the size of the alkyl chain of cation orthe IL concentration at 85 �C. In this way, the excellent protectiveeffect of ILs based on [NTf2] on enzyme activity against denaturingagents (e.g. temperature, methanol, etc.) has been reported[18,31,32], and should be considered as an added value of ILs forenzymatic biodiesel synthesis. The observed improvements for li-pase-catalyzed synthesis of biodiesel in ILs reaction media by theincrease in the alkyl chain length of the cation (e.g. [Emim],

3466 P. Lozano et al. / Fuel 90 (2011) 3461–3467

[Bmim], [Hmim] and [Omim]) have been also reported by otherauthors, the best result (up to 80% biodiesel yield in 20 h) being ob-tained for the [Omim][NTf2] case [17].

3.3. Continuous enzymatic reactor for biodiesel synthesis inhydrophobic ILs/scCO2 biphasic systems

Biphasic catalytic systems based on a combination of ILs andscCO2 with enzymes may represent the most important ‘‘arsenal’’of green tools for developing integrally clean chemical processesof industrial interest in the near future [28,29]. In this context, abiocatalytic packed bed reactor, containing immobilized lipaseparticles coated with ILs, was developed to carry out the continu-ous synthesis of biodiesel in scCO2 phase (see Fig. 2). The systemoperated as a biphasic reactor, the substrates being transportedby the scCO2 phase to the biocatalyst microenvironment acrossthe IL layer, and the products returning to the supercritical phase(see Fig. 1), in order to produce directly pure biodiesel. Five differ-ent ILs, i.e. [C18mim][NTf2], [C18mim][PF6], [C18mim][BF4],[C14mim][NTf2], [C14mim][PF6], were assayed, and each Nov-ozym-IL system was continuously tested for 8 h cycles (seeFig. 5). It was reported how the solubility of triolein in scCO2 isenhanced by increasing pressure (150–350 bar), temperature(40–80 �C) as the scCO2 density increases, and by the presence ofa low amount of methanol (up to 5% v/v) as cosolvent [33]. As afunction of these data, the operational conditions (i.e. scCO2 at60 �C and 180 bar, 5.1 and 10.2 lmol triolein min�1 and244.4 lmol methanol min�1) were selected to provide the fullsolubilisation of substrates at the reactor inlet. As can be seen inFig. 5, the proposed reactor was suitable for the continuous synthe-sis of biodiesel, resulting in a full conversion of substrate in fourNovozym-IL systems (i.e. [C18mim][NTf2], [C18mim][PF6], [C14mim][NTf2] and [C14mim][PF6]) during the first five operation cycles,where the triolein inlet flow was 5.1 lmol min�1. The coating ofthe enzyme with [C18mim][BF4] provided the worst results, thebiodiesel yield being reduced up to 31% after three operational cy-cles, probably because of the hydrophilic character of [BF4] anion,which provided a poor ability of this IL to dissolve both trioleinand methyl oleate (see ESI), and so, to transport triolein from thescCO2 phase to the active site of the enzyme. The use of the immo-bilized enzyme without coating with IL showed a full trioleintransformation to methyl oleate during the first three operational

Operation Cycles2 4 6 8 10 12

Bio

dies

el Y

ield

(%)

0

20

40

60

80

100A B

Fig. 5. Operational stability for continuous biodiesel synthesis catalyzed byNovozym 435 coated with IL ([C18mim][NTf2] (N), [C18mim][PF6] (j),[C18mim][BF4] (�), [C14mim][NTf2] (4), [C14mim][PF6] (h) and none (s), respec-tively) in scCO2 at 60 �C and 180 bar, and at 5.1 lmol triolein min�1 and244.4 lmol methanol min�1 (A), or 10.2 lmol triolein min�1 and 244.4 lmol meth-anol min�1 (B) mass-flow rates, respectively. Each point corresponds to the middlevalue of all samples obtained during each 4 h cycle.

cycles, then being continuously reduced. This decrease in the enzy-matic activity could be attributed to the reported denaturing effectof scCO2 on enzymes [28], and/or by the low efficiency of thehydrophobic scCO2 phase to transport the hydrophilic by-productglycerol [34], which could be retained around the enzyme micro-environment leading to the continuous biocatalyst poisoning bythe prevention of the entry of new triolein substrate molecules[3]. However, the increase by two-fold of triolein inlet flow (from5.1 to 10.2 lmol min�1) led to a fall in the biodiesel yield for allcases. The best results were obtained for the Novozym-[C18mim][PF6] system, where a high biodiesel yield (82%) was ob-tained after 12 cycles of continuous operation. These results couldbe explained as a function of the phase behaviour of IL/triolein/methanol and IL/methyl oleate mixtures, as well as, by its influenceon the enzyme activity. In this context, the excellent ability of[C18mim][NTf2] to dissolve triolein could clearly be in the reasonfor the highest biodiesel yield obtained during the first six cyclesin scCO2 conditions. However, the full miscibility of this IL withboth the triolein substrate and the methyl oleate product (seeFig. 3 and ESI) could also be involved in the continuous activity de-cay observed at long operation cycles. Thus, an excess in the trio-lein inlet flow or the continuous release of biodiesel productfrom the enzyme particle to the scCO2 flow may dissolve the pro-tective IL shell, enhancing the continuous enzyme deactivation. Infact, peaks corresponding to this IL were observed in GC analysis ofthe outlet product (see ESI). However, product samples obtainedfrom the Novozym-[C18mim][PF6] outlet flow did not show peakscorresponding to the release of this IL, which could be related withthe improved operational stability showed by the immobilized en-zyme coated with this IL. The poor miscibility of [C18mim][PF6]with triolein or methyl oleate with respect to [C18mim][NTf2], to-gether the high melting point of this IL (82 �C) could be involvedin the preservation of the IL shell around the enzyme particles un-der continuous scCO2 flow at 180 bar and 60 �C. Additionally, itwas reported that ILs exhibits a melting point depression (evenexceeding DTm of 100 �C) in scCO2 [35], and this fact could be afurther indication towards the better stabilization of the enzymeby the [C18mim][PF6] shell in liquid state, thus protecting the lipasefrom methanol and scCO2 induced deactivation. The ability ofscCO2 to extract hydrophobic compounds (e.g. olefins) from ILshas been widely described, where the partitioning behaviour of or-ganic compounds between both IL an scCO2 phases is clearlydependent on the nature of the IL and the supercritical conditions.The use of scCO2 (35 �C, 85 bar) to discontinuously extract butyloleate from the IL methyltrioctylammonium trifluoroacetate hasbeen reported as a later step of the enzymatic butanolyis of trio-lein. The best results for this enzymatic reaction (90% butyl oleateyield) were obtained at 80% IL content, but data about the effi-ciency for the butyl oleate extraction with scCO2 were not provided[36].

4. Conclusions

The suitability of enzymes to continuously catalyzed biodieselsynthesis under green non-aqueous reaction media, such as ILand scCO2, has been demonstrated. The unique properties of longchain ILs, providing a microenvironment for enzyme-catalyzedreactions, led to a clear improvement in the efficiency for biotrans-formation of vegetable oil to biodiesel. Additionally, the IL anionalso plays an important role on this efficiency, which is improvedby the increase in its hydrophobicity (i.e. [NTf2] > [PF6] > [BF4]).Accordingly, the enzymatic synthesis of biodiesel in IL/scCO2 bi-phasic systems under continuous operation can easily be achievedbecause of the suitability of the supercritical phase to efficientlytransport the biodiesel product. However, further studies towards

P. Lozano et al. / Fuel 90 (2011) 3461–3467 3467

designing to provide a permanent coating of enzyme with hydro-phobic ILs (e.g. by using covalently supported ionic liquid phases,(SILPs) [37]) should improve the operational performance of thesebiocatalytic systems in continuous processes for biodiesel synthe-sis in scCO2. This approach may bring to the biofuel industry a clearstrategy for developing continuous integral sustainable processesfor biodiesel synthesis.

Acknowledgements

This work was partially supported by CICYT (CTQ2008-00877)and SENECA Foundation (08616/PI/08) Grants. We thank RamiroMartinez (Novozymes España. S.A.) for the gift of Novozym 435�.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.fuel.2011.06.008.

References

[1] Atadashi IM, Aroua MK, Aziz AA. Renew Energy 2011;36:437–43.[2] Vyas AP, Verma JL, Subrahmanyam N. Fuel 2010;89:1–9.[3] Bajaj A, Lohan P, Jha PN, Mehrotra R. J Mol Catal B Enzym 2010;62:9–14.[4] Rodrigues RC, Fernandez-Lafuente R. J Mol Catal B Enzym 2010;64:1–22.[5] Tan T, Lu J, Nie K, Deng L, Wang F. Biotechnol Adv 2010;28:628–34.[6] Dupont J, Suarez PAZ, Meneghetti MR. Energy Environ Sci 2009;2:1258–65.[7] Shimada Y, Watanabe Y, Samukawa T, Sugihara A, Noda H, Fukuda H, et al. J

Am Oil Chem Soc 1999;76:789–93.[8] Lu J, Nie K, Xie F, Wang F, Tan T. Process Biochem 2007;42:1367–70.[9] Luo Y, Wang G, Ma YS, Wei DZ. J Chem Technol Biotechnol 2006;81:1846–8.

[10] Du W, Xu Y, Liu D, Zeng J. J Mol Catal B Enzym 2004;30:125–9.

[11] Royon D, Daz M, Ellenrieder G, Locatelli S. Bioresource Technol2007;98:648–53.

[12] Liu Y, Yan YJ, Hu F, Yao AN, Wang ZC, Wei FX. AIChE J 2010;56:1659–65.[13] Oliveira C, Rosa MF. J Am Oil Chem Soc 2006;83:21–5.[14] Varma MN, Madras G. Ind Eng Chem Res 2007;46:1–6.[15] Varma MN, Deshpande PA, Madras G. Fuel 2010;89:1641–6.[16] Shunitha S, Kanjilal S, Reddy PS, Prasad RBN. Biotechnol Lett 2007;29:1881–5.[17] Ha SH, Lan MN, Lee SH, Hwang SM, Koo YM. Enzyme Microb Technol

2007;41:480–3.[18] Gamba M, Lapis AAM, Dupont J. Adv Synth Catal 2008;350:160–4.[19] Arai S, Nakashima K, Tanino T, Ogino C, Kondo A, Fukuda H. Enzyme Microb

Technol 2010;46:51–5.[20] Silva WSD, Llapis AAM, Suarez PAZ, Neto BAD. J Mol Catal B Enzym

2011;68:98–103.[21] Chen B, Guo Z, Tan T, Xu X. Biotechnol Bioeng 2008;99:18–29.[22] van Rantwijk F, Sheldon RA. Chem Rev 2007;107:2757–85.[23] Lozano P, de Diego T, Iborra JL, Vaultier M. Patent PTC/WO 2010/057996 A6;

2010.[24] Lozano P, Bernal JM, Piamtongkam R, Fetzer D, Vaultier M. ChemSusChem

2010;3:1359–63.[25] de Diego T, Manjón A, Lozano P, Vaultier M, Iborra JL. Green Chem

2011;13:444–51.[26] Abe Y, Kude K, Hayase S, Kawatsura M, Tsunashima K, Itoh T. J Mol Catal B

Enzym 2008;51:81–5.[27] Ha SH, Lan MN, Koo YM. Enzyme Microb Technol 2010;47:6–10.[28] Lozano P. Green Chem 2010;12:555–69.[29] Lozano P, De Diego T, Mira C, Montague K, Vaultier M, Iborra JL. Green Chem

2009;11:538–42.[30] Dupont J. J Braz Chem Soc 2004;15:341–50.[31] de Diego T, Lozano P, Gmouh S, Vaultier M, Iborra JL. Biomacromolecules

2005;6:1457–64.[32] Micaelo NM, Soares CM. J Phys Chem B 2008;112:2566–72.[33] de Lucas A, Gracia I, Rincon J, Garcia MT. Ind Eng Chem Res 2007;46:5061–6.[34] Sovova H, Jez J, Khachaturyan M. Fluid Phase Equilib 1997;137:185–91.[35] Roth M. J Chromatogr A 2009;1216:1861–80.[36] Miyawaki O, Tatsumo M. J Biosci Bioeng 2008;105:61–4.[37] Lozano P, Garcia-Verdugo E, Piamtongkam R, Karbass N, de Diego T, Burguete

MI, et al. Adv Synth Catal 2007;349:1077–84.